On Thu, 25.May.23 10:27:48 +0000, mutt-dev-requ...@mutt.org wrote: > > Message: 7 > Date: Thu, 25 May 2023 12:25:42 +0200 > From: Ludolf Holzheid <lholzh...@bihl-wiedemann.de> > To: mutt-dev@mutt.org > Subject: Re: What is a Message-id? > Message-ID: <20230525102542.ga10...@mail.bihl-wiedemann.de> > Content-Type: text/plain; charset=utf-8 > > On Thu, 2023-05-25 10:59:38 +0100, Ian Collier wrote: > > [..] > > > > Yes, the original MUA is 100% wrong on that point. The question here is > > whether Mutt should compensate for it (see "Robustness principle"). > > But for the scenario Olaf described, Mutt comes only into play when > the child has already fallen into the well: > > ? The first MUA produces a message-id header without angle brackets. > ? The second MUA adds angle brackets to meet the standard.
I find it very likely that the second MUA would consider both messages part of the same thread. > ? Mutt (the third MUA) doesn't compare the (now different) > message-ids equal. Here I tend to disagree. While the field syntax requires angle brackets *around* the message-id, they are not part of the message-id value itself. So Mutt should consider them equal. > The only thing Mutt that could do about it, is relaxing the > comparison from ?equal or not? to ?maybe somewhat equal enough?. Would it bring a huge risk if the angle brackets are stripped from all message-ids? Then the comparision can remain "equal" and by accepting the header without brackets Mutt can be liberal in what is received and produce a valid header in a response, exactly what the MUA 2 did. Regards Andrej
